Unless you are very close to the subject the perspective change won't be much given how far apart the sensors would be. Perspective changes really become an issue if doing very close photography, especially macro where you also run into things like focus breathing when doing focus stacking. If you are at least several feet away from the subject the perspective change shouldn't be a big issue and at larger distances becomes pointless. It may introduce some additional softness in high detail areas but with all of the other issues around image quality, tiny sensors, small pixel pitches and diffraction limits of ideal lenses it likely won't make that much of a difference. I do wonder if the perspective errors at distances of 2-3 meters would be more than the chromatic aberrations and diffraction of those little 3-4 mm lenses though. It looks like human vision can detect
parallax differences down to about 30 arc seconds which seems like a good value to use since our eyes are farther apart than these tiny cellphone cameras so it would be a gross over estimate of the parallax they would experience. As most cell phone lenses are around a 28mm full frame equivalent that means they provide about a
70 degree field of view. So 70 degrees times 60 arc minutes per degree times 2 (30 arc seconds) gives us 8400 30 arc second units. This means to fully resolve parallax differences that people can see we would need to have a 70.5 mega pixel camera with a non diffraction limited 28mm lens in front of it. However a 28mm equivalent lens in front of a modern cellphone sized sensor really means the camera has about 4000 1 arc minute wide pixels. So if you don't notice parallax effects with your eyes your cellphone's quad cameras really won't.
In looking at the resolving power of lenses most cellphones unless they have a very wide open lens, think f/1.2, they are diffraction limited. Take the iPhone's 10XS's f/1.8 lens and 12 MP sensor. Under ideal circumstances the airy disc produced by a point source of light (2.4um diameter) will cover just under a 2x2 grid of pixels (pixel pitch of 1.4 um) for it's "normal" lens, and the f/2.4 "telephoto" lens would produce an airy disc that covers about a 3x3 grid of pixels under ideal circumstances. So using this we can see that diffraction has an effect about 3 to 4 times as greater, affecting 1.5 to 2 arc minutes, than the minimum human eye noticeable parallax would on image sharpness. Also very wide lenses like these tend to be on the soft side of things which only further degrades the resolving power. Here we have only been looking at image sharpness which is different from noise but still affects overall image quality.
On the noise front you won't see an exact halving of the noise but it will be very close. Even under ideal circumstances where one has a uniform colored uniformly illuminated shot the noise reduction would be less than half but is close enough to say so. This is because there will be systematic noise that is introduced into the system in addition to the truly random noise from the flow of photons to the sensor. The random noise will be halved but the noise introduced by the AtoD conversion, the signal amps, and a non uniform ideal sensor will still be present. If one wants to, it is possible to subtract out a fair amount of this systematic noise. Doing so requires dark and bias frames to be taken and is common practice in astrophotography.